Success Story: HPC as Industry Enabler
NCC presenting the success story
Leonardo is a third party in the Italian National Competence Centre EuroCC Italy, established by the EuroCC project. Leonardo experts employ their HPC competences, acquired through experience on the Davinci-1 supercomputer, to provide the Italian companies, in particular from the Liguria region, with technical support and training in HPC/AI/HPDA field.
The HPC is indeed a central infrastructure in supporting companies to innovate and improve their competitiveness in the respective business sector. Actually, there are many companies that may profit from HPC technologies and methodologies: from the SMEs that could change and speed up their algorithms, to the large companies that could upgrade existing processes and create fertile ground for the creation of completely innovative solutions.
This result confirms the real possibility of exploiting HPC platforms in cases of pandemics. Also, during the project precious web platforms to support the global research community with bioinformatics and simulation tools were deployed, including MEDIATE – MolEcular DockIng AT home (https://mediate.exscalate4cov.eu/) – will give free access to the largest database available today on the Sars-CoV-2 Virus both from a structural (three-dimensional structures) and functional (proteins interacting with human cells) point of view. In https://viralseq.exscalate4cov.eu/, the viral mutations retrieved using genomic data from public repositories (i.e. GISAID, EMBL COVID-19 data portal) are mapped and analysed in their 3D structural context to investigate their impact in terms of host-immune interaction, ligand/substrate/drug binding sites and SDPs; The other web portal is https://spikemutants.exscalate4cov.eu/, that aims to provide the scientific community with structural information on emerging variants involving the protein sequence of the Sars-CoV-2 Spike protein; The other impressive release will be http://SCoV2-MD.org, a website that include MD trajectories retrieved worldwide and analysis tools to better understand the dynamic behaviour of viral proteins.
Exscalate4CoV, using a unique combination of high performance computing power and AI with biological processing, brings together 18 partners and further 15 associated members. This includes supercomputing centres in Italy, Spain and Germany, large research centres, pharmaceutical companies and biological institutes from across Europe. The platform has around 120 Petaflops computing power, allowing research into the behaviors of molecules with the aim of identifying an effective treatment against coronavirus. The project’s chemical library is constantly growing thanks to agreements with newly associated pharmaceutical companies.
The consortium has virtually tested 400 000 molecules using its supercomputers. 7 000 molecules were preselected and further tested “in vitro”. Raloxifene emerged as a promising molecule: according to the project, it is effective in blocking the replication of the virus in cells, and could thus hold up the progression of the disease. Researchers have indicated that its advantages include its high patient tolerability, safety and established toxicological profile. The consortium is discussing with the European Medicines Agency how to advance to clinical trials to evaluate the new potential use for Raloxifene. If successful, the drug could be quickly made available in high volumes and at low cost.
Leonardo as EuroCC Competence Center:
Leonardo is a third party in the Italian National Competence Centre EuroCC Italy, established by the EuroCC project and aims to provide SMEs with technical support and expertise by Leonardo’s HPC – researchers and experts.
As a leading technology company in Italy, Leonardo supports the national ecosystem trying to enable and encourage local companies with the adoption of HPC technologies to gain a competitive advantage in the business sector. Thanks to the collaboration with the technological hub of SIIT in Genoa, many training courses focusing on HPC have been delivered and targeted to local SMEs willing to gain closer collaboration with the National Competence Center. The training courses range from the management of HPC infrastructure, to modeling and parallelization of code with GPUs, to the creation of Big Data systems based on Cloud Computing, to the development and training of Artificial Intelligence models using HPC.
Leonardo is strongly supported by CINECA, with its long-standing experience in HPC technology transfer and leader of EuroCC Italy, for disseminating HPC technologies and offering specialized courses.
Leonardo as HPC User:
Considering the wide portfolio of products and services built and developed by Leonardo, HPC has proven to be useful in different scenarios and sectors. Those, HPC has a dual function: on one hand, it provides a large proprietary computing power, by removing limits on the processing power used by the internal algorithms and codes; on the other, it allows centralized computing capacity for the company which lead to optimize the expertise of engineers and researchers. Within the company, HPC and cloud computing resulted in designing and developing innovative solutions in the fields of Big Data, Artificial Intelligence, Simulation of complex systems and Optimization of parallel code. More, “Digital Twin” will be one of the main reality for future Leonardo HPC -related activities.
The term ‘Digital Twin’ arised from the the engineering field, and stands for a computer program, fed with data collected from a real system, and able to represent in a synthetic but accurate way (often through visualizations with 3D models, graphs, curves and dashboards) the overall status of the real twin. Simplifying, we could say that the digital twin is the equivalent of a control unit of the real twin, implemented within a software and that can work even without the controlled system. Recently, the concept of Digital Twin is taking on new meanings in the sense of a holistic digital model of a real system, or a virtual representation of it (always within a computer program) that replicates the state and changes in state, thanks to the combined use of data, simulations and artificial intelligence. The holistic model as an extension of the Digital Twin is an incredibly powerful tool since it allows for predictive capability. It is largely enabled by the computing and data analysis power now available in supercomputers or in the cloud. The power is such to be able to calculate very complex but accurate numerical models, able to meet the increasing need to predict the behavior of a system under different operating conditions, be it a car an airplane, a ship, an industrial plant up to the human body and the whole earth (in the latter two cases we are still talking about research projects). The availability of an accurate and predictive virtual twin of the processes within Leonardo is also essential to predict the effect of a change of state, intended or unintended (e.g. due to altered environmental conditions), to avoid malfunctions, to reduce production and operating costs, by virtue of preventive actions, to make «what-if», evaluations, to train operators, etc…
The key element in developing a digital twin is the software. Software is not a single program; conversely, it implements complex multi-component (engine, structure, air, water, etc…) and multi-scale models (metal, metal components, complete aircraft, a fleet of airplanes, etc… ), which are evaluated in a coupled fashion, i.e., whose values affect each other. Hence, a special function describing the internal state of the system and keeping all components synchronized, is strongly required. To properly define a Digital Twin, this function must maintain synchronization also with the values of the sensors within the real twin.
There are the other two elements very relevant for digital twin: the data, which may be collected from sensors or simulated, and the numerical models needed to simulate the behavior of the various system components. The models can be based on the first principles, i.e. the constitutive equations that describe the behavior of the subsystem/component, or can be ‘data driven’, when data collected by the sensors are used to define an implicit model of the system behavior, through procedures ranging from simple interpolation to artificial intelligence. In this way, procedures computationally very expensive are able to faithfully replicate the states of a subsystem when the input data change (in this case, the AI model reproduces the load curve without any equation!).
Therefore, the realization of a Digital Twin is a very complex work and requires many elements, as a large amount of data, an adequate computational infrastructure, specific software and, above all, transversal competences, ranging from experts in real systems, to process and design engineers, to computer scientists for software writing and data management, to mathematicians or physicisst for models, to computational infrastructure experts.
Leonardo experiments this innovation improvement thanks to HPC- activities on the Davinci-1 supercompuer, which ensured the computational power needed to develop and execute the software, manage and process the massive amount of data, and compute the solutions to the various mathematical and artificial intelligence models needed to create the digital twin.
SUCCESS STORY # HIGHLIGHTS:
- Digital Twin
- Industry Enabler
- Defence and Security
- Industrial Processes
This project has received funding from the European High-Performance Computing Joint Undertaking (JU) under grant agreement No 951732. The JU receives support from the European Union’s Horizon 2020 research and innovation program and Germany, Bulgaria, Austria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, Greece, Hungary, Ireland, Italy, Lithuania, Latvia, Poland, Portugal, Romania, Slovenia, Spain, Sweden, the United Kingdom, France, the Netherlands, Belgium, Luxembourg, Slovakia, Norway, Switzerland, Turkey, Republic of North Macedonia, Iceland, Montenegro